1.4.0

Major change in how End-of-Line is handled, supporting Unix+Windows encoding.

Previously, it was relying on detection, but now it just changes the encoding dynamically, to Unix-like \n, which simplifies processing and makes it faster.

In addition, the library now changes Unix-encoding inside of special/copyright comment blocks to Windows encoding, to make sure that special-block formatting created on Linux looks the same on Windows.